Enterprises know AI-generated code is vulnerable; they’re shipping it anyway

Checkmarx surveyed 2,350 CISOs, AppSec managers, and developers and found that nearly half of production code is AI-generated and much of codebases rely on open-source components; the report warns that powerful models such as Anthropic's Mythos can accelerate discovery and weaponization of vulnerabilities from months to minutes, meaning traditional security approaches may be insufficient.